Integrating Clustering and Gradient Boosting Tree for eDNA - based Early Detection of Invasive Species in Ship Ballast

Environmental DNA (eDNA) technology has great potential for detecting alien species in ships' ballast water and sediments, and can help with early detection of invasions. In this study, this technique was applied. Two primers were designed and combined with full-length diversity sequencing and microbial community diversity sequencing. In the analysis of microbial community diversity, K-means clustering and hierarchical clustering algorithms are used to optimize operational classification unit (OTU) clustering. In the process of full-length diversity analysis, several machine learning models were compared, and the gradient boosting tree model was selected for species classification optimization. The study found that the dominant species in Marine sediments are diatomata (about 9%) and chlorophyta (about 30%). These results provide strong support for the management of species in ballast water and sediments, the prevention and control of invasions of dominant species, and the warning of high-risk ships and shipping routes.